Identification and Avoidance of Multiple Reflections and Posterior Echo Enhancement using Robotic Echocardiography

In this study, we propose a method to identify and avoid multiple reflections and posterior echo enhancement by robotic echocardiography, with the aim of improving the search accuracy of the basic examination cross-section acquisition system developed by the authors. We developed a discrimination algorithm based on the characteristics that a layered false image appears at integer multiples of the distance from the probe to the real image for multiple reflections, and a bar-shaped false image appears behind the pericardium for posterior echo enhancement. We then constructed a method to avoid multiple reflections and posterior echo enhancement by using the proposed identification method. When the left ventricular wall of the closed loop was not detected because of multiple reflections, the posture of the probe was adjusted, and the search was performed again. When the center of gravity of the detected ventricular septum was included inside the contour of the posterior echo enhancement, the detected ventricular septum was judged as a false positive. The evaluation of the identification method was performed on 300 images, of which 100 images each showed multiple reflections and posterior echo enhancement and 100 images showed no artifacts. The results showed that multiple reflections and posterior echo enhancement were detected with an F-score of 82.1 % and 90.7 %, respectively, suggesting the validity of the identification method. The results showed that the robot was able to detect the left ventricular wall and ventricular septum in a closed loop with 21.3 % and 62.7 % accuracy, respectively, suggesting the effectiveness of the avoidance method.